The latest edition of The Gretel Epoch is live. 🚀 ⌛ This cycle, we explore the power of synthetic code, new strategies for building privacy-compliant chatbots, and take time to reflect. 🔁 Let's GO. 👇🏼 #SyntheticData #GenAI #Privacy
Gretel
Software Development
Palo Alto, California 19,889 followers
The synthetic data platform purpose-built for Generative AI
About us
Gretel is solving the data bottleneck problem for AI scientists, developers, and data scientists by providing them with safe, fast, and easy access to data without compromising on accuracy or privacy. Designed by developers for developers, Gretel’s APIs make it easy to generate anonymized and safe synthetic data so you can preserve privacy and innovate faster. You can learn more about synthetic data from Gretel's engineers, data scientists, and AI research team on our blog: https://gretel.ai/blog
- Website
-
https://gretel.ai
External link for Gretel
- Industry
- Software Development
- Company size
- 51-200 employees
- Headquarters
- Palo Alto, California
- Type
- Privately Held
- Founded
- 2020
- Specialties
- Generative AI, Synthetic Data, Privacy, AI, and Deep Learning
Products
The Developer Stack for Synthetic Data.
Data Privacy Management Software
Synthetic data that’s as good, or even better than the data you have. Or don’t have. Create and share data with the best-in-class accuracy and privacy guarantees – on demand.
Locations
-
Primary
Palo Alto, California, US
-
San Diego, California 92122, US
Employees at Gretel
-
Yamini Kagal
-
Bryan Zimmer
Head of Security at Gretel.ai
-
Maarten Van Segbroeck, Ph.D.
Head of Applied Science at Gretel.ai l ex-Amazon
-
Jeff Gilling
Passionate about #technology solutions that deliver better experiences #ageing #agetech #healthtech #fintech #agtech #startups #VC #socent…
Updates
-
Training and deploying LLMs presents unique challenges in managing data privacy risks—challenges that traditional de-identification methods can't fully address. That's why Gretel advocates for synthetic data as a privacy-preserving solution and why we've been developing advanced tools to quantify these risks. We’ve recently added two new enterprise-grade metrics for evaluating synthetic tabular data privacy risks in real-world adversarial scenarios: 🛡️ Membership Inference Attack (MIA): Simulates an attacker attempting to determine whether a specific record from real data was used to train the synthetic model. Essentially, it measures if synthetic data "leaks" any real records. 🛡️ Attribute Inference Attack (AIA): Focuses on inferring unknown attributes of synthetic data using known portions of real data, highlighting potential vulnerabilities where sensitive information could be exposed. The attached graphic provides a visual overview of how these attacks work. To dive deeper into the science of measuring privacy risk, check out our applied research blog: https://lnkd.in/errUVV7F #Privacy #AI #DataGovernance
-
Honored to have been amongst the first participants to complete the newly announced Google Cloud ISV Startup Springboard program. It helped us accelerate growth and get set up for co-sell with Google Cloud. Read all about it. https://lnkd.in/gyJM6gPH #GoogleCloud #GoogleCloudSpringboard #ISV #startups
See how AI startups are accelerating their growth with Google Cloud | Google Cloud Blog
cloud.google.com
-
Gretel reposted this
As part of our deepening partnership with Google Cloud we are making it easier than ever to generate synthetic data natively within BigQuery using Gretel. You can create privacy-preserving synthetic datasets while maintaining the original schema and structure, with no loss to downstream utility, right inside your existing workflows. Key Benefits: - Enhance Privacy: Generate synthetic data that complies with regulations like GDPR and CCPA, and future proof your data against any regulation. - Boost Data Access: Safely share and analyze data without exposing sensitive information. Make data easy, fast, and safe to work with. - Accelerate Development: Use synthetic data for testing and model training, free from production data risks. Accelerate your transition from training to inference and deliver real value for your business. This partnership empowers teams to innovate faster, securely, and at scale.
Create synthetic data with Gretel in BigQuery | Google Cloud Blog
cloud.google.com
-
Gretel reposted this
We're proud to announce a new product integration with Gretel, which enables #BigQuery users on Google Cloud to generate and use privacy-safe synthetic data for model training, data sharing and other use cases, all via BigQuery DataFrames! Check out the first part of a two part blog post series below. The second post will be a more technical "how to" guide, coming soon. Thanks to all of our friends Gretel for the continued partnership and collaboration! cc: Yasmeen Ahmad, Gerrit Kazmaier, Laurel Krieger, Ali Arsanjani, PhD, Firat Tekiner, Wu Jiaxun, Naveen Punjabi, Jean Ji, Ali Golshan, Alexander Watson, Rebecca Kao, John Myers https://lnkd.in/g8tmKRF5
Create synthetic data with Gretel in BigQuery | Google Cloud Blog
cloud.google.com
-
Gretel reposted this
Big day for #Startups and Google Cloud! Today we announced the new ISV Startup Springboard program to help AI and cyber-security startups accelerate growth and get co-sell ready! Read all about it, hear from some of the great partners who have already benefited from the program including Baseten Dataloop AI Galileo 🔭 Gretel HumanFirst Roboflow and register your interest today! https://lnkd.in/ecUf8iHX Ritika Suri Stephen Orban Ashwin Karuhatty Vince Bryant Tai Conley Kashaf Mazhar Amy Bray Karen Sigman Mary Lynn Stryker Judy Wu
-
Gretel reposted this
#COLM Spotlight paper: “GPQA: A Graduate-Level Google-Proof Q&A Benchmark“ Remember the incredible o1 results pointing to PhD-level intelligence on challenging tasks in physics, chemistry, and biology? Well these are based on a GPQA dataset that is less than a year old (!) and took quite a bit of effort to put together by folks from New York University, Anthropic and Cohere. Today I met David Rein, one of the people behind the GPQA dataset. David openly pondered “Is GPQA dead on arrival”? – the short answer is no! – while the progress on the benchmark has been nothing short of mind-boggling (see the red curve in the photo below), GPQA can be used to simulate supervising superhuman AI systems – this lets us develop scalable oversight methods that let us more easily evaluate new models with substantially deeper expertise – this is very much needed to make benchmarking more sustainable, given the human and capital cost of evals like GPQA More broadly, much more investment and progress is needed on the evaluation side to keep up with the rapid pace of progress in model capabilities. And guess what? Synthetic data, including data from Gretel, is going to be a huge part of it all!
-
Gretel reposted this
Are you a #productmarketing player/coach? If so, come join Gretel the #syntheticdata platform purpose-built for #GenAI to lead our product positioning & messaging on our marketing team. Gretel was recently named one of LinkedIn’s Top U.S. Startups! https://lnkd.in/gvtT2c7R #hiring #remoterole based in US/Canada
Head of Product Marketing
jobs.ashbyhq.com
-
Gretel reposted this
Super excited about the inaugural Conference on Language Modeling (#COLM) in Philadelphia next week! So many interesting topics to be discussed, including: -- All about #data: pre-training data, alignment data, and #synthetic data --- via manual or algorithmic analysis, curation, and generation -- All about #safety: security, #privacy, misinformation, adversarial attacks and defenses -- All about #evaluation: benchmarks, simulation environments, scalable oversight, evaluation protocols and metrics, human and/or machine evaluation -- All about #alignment: fine-tuning, instruction-tuning, reinforcement learning (with human feedback), prompt tuning, and in-context alignment Attending? Let me know, and let's connect in person!
-
Data privacy is critical, especially when dealing with sensitive customer information. Gretel's Workflows streamline multi-step synthetic data generation, helping organizations securely transform and synthesize data at scale. In our latest demo, we connect to external data sources, redact PII with Transform, generate privacy-safe synthetic data using Navigator Fine-Tuning, and store the results in cloud storage. This end-to-end process ensures that sensitive data is both anonymized and synthesized, making it safe to share while maintaining privacy and data integrity. Watch demo: https://lnkd.in/eQKC7zYe #AI #SyntheticData #DataSecurity